基于NILM的电动自行车室内充电监测  

Non⁃intrusive load monitoring(NILM)for indoor electric bicycle charging monitoring

作　　者：陈志高 李琰 徐天奇 CHEN Zhigao;LI Yan;XU Tianqi(School of Electrical and Information Engineering,Yunnan Minzu University,Kunming 650504,China;Yunnan Key Laboratory of Unmanned Autonomous System,Yunnan Minzu University,Kunming 650504,China;Key Laboratory of Electric Power Information and Physical Fusion System of Universities in Yunnan Province,Yunnan Minzu University,Kunming 650504,China)

机构地区：[1]云南民族大学电气信息工程学院,云南昆明650504 [2]云南民族大学云南省无人自主系统重点实验室,云南昆明650504 [3]云南民族大学云南省高校电力信息物理融合系统重点实验室,云南昆明650504

出　　处：《现代电子技术》2025年第9期167-172,共6页Modern Electronics Technique

基　　金：国家自然科学基金项目(62062068);云南省青年学术和技术带头人计划(202305AC160077)。

摘　　要：电动自行车室内充电导致的火灾事故频发,监测其室内充电行为成为一大难题。文中通过研究各类品牌与型号电动自行车充电时共同的负荷特性,提出基于小波事件检测和两阶段特征选择的非侵入式电动自行车室内充电监测系统。首先,利用多层小波变换对信号发生突变时的敏感性,基于电动自行车充电时负荷的暂态特性捕捉充电事件,提高检测精度并降低计算量。然后,利用其稳态特性进行两阶段特征选择:第一阶段使用MDMR过滤器对14种负荷特征的重要性进行排序;第二阶段采用OCSVM作为包装器筛选最优特征子集。实际用户监测实验验证了该方法能够做到快速响应和高精度识别并且计算成本低,为电动自行车室内充电监测提供了一种经济高效的解决方案。The increase in fires caused by indoor charging of electric bicycles(EBs)has made effective monitoring a challenging issue.By analyzing common characteristics of charging loads of the EBs with various brands and models,a non-intrusive real-time EB indoor charging monitoring system based on wavelet event detection and a two-stage feature selection process is proposed.In the system,the sensitivity of multi-scale wavelet transform to signal mutation is used to detect EB charging events based on transient characteristics when the load connects to the circuit,improving detection accuracy and reducing computational load.Then,on the basis of steady-state characteristics during operation,a two-stage feature selection is employed.In the first stage,the MDMR(maximum discrimination and minimum redundancy)filter is used to rank the importance of 14 load features.In the second stage,the OCSVM(one-class support vector machine)is used as a wrapper to select the optimal feature subsets.The actual user monitoring experiments verify that the proposed method can achieve fast response,high precision identification and low calculation cost,so it provides an economical and efficient solution for indoor charging monitoring of EBs.

关 键 词：非侵入式负荷监测 电动自行车 室内充电 小波变换 特征选择 一类支持向量机 

分 类 号：TN609-34[电子电信—电路与系统] TM930.1[电气工程—电力电子与电力传动]